Research in graduate school is about reading, learning, experimenting, and writing about what you've learned. Writing is how to communicate what you've done in your research, what you know now, and what you'd like to do. Some of us are fascinated when it comes to the writing part of the research process. Others prefer the experimentation stage of research and are reluctant to write about the results. In any case, you're guaranteed a large amount of writing during grad school, including class project reports, conference papers, journal articles, and funding proposals.
Four years of grad school, a few conference/journal manuscripts, and several class project reports later, here's a list of suggestions that I use when writing. There are tons of online resources authored by experts to help improve your writing skills—I'm providing the student perspective of what it means to write day after day, whether you love the writing process or you're a reluctant writer.
How to get started
By the time you sit in front of the computer, you should have some idea of how you want to organize your thoughts. If typing isn’t your best way to get inspired, consider working on an outline on paper before starting.
Separate writing from editing. It takes a lot of effort to put sentences in the right order and you might lose your line of thought by the time you're done. Instead, "close your eyes" (a professor once suggested covering your screen entirely) and write everything you have in your head. Edit afterwards!
How to write the abstract
A reader will decide if they're interested in your paper based on the abstract (and the title, of course). Therefore, your job is to convince the reader to stay with you. You need to be clear about what you'll be talking about while conveying the message, "Read more if you want to hear the details." A paper is a story told multiple times and the abstract is the first time you tell it.
Try to include several key words that will catch the reader's eye. For example, "We propose two approaches that could be used by President Obama regarding the new situation with Cuba" is more provocative than "We discuss what could be done next regarding the situation with Cuba."
How to write the conclusion
Simply put, the conclusion is your final opportunity to convince the reviewer that your paper is worth publishing or the reader that your argument is a strong one. Start with a strong statement. "We proposed a novel technique for cloud-based security that increases the reliability of the system by 20% compared to previous studies" conveys novelty and successful results, and shows readers who read only the conclusion that your topic is cloud-based security. This statement is much stronger than, "In this paper we have investigated the problem of cloud-based security." After the strong opening, you can briefly summarize your work and describe its advantages.
Don’t tell new stories in the conclusion. It's okay to describe future work, but you shouldn't introduce concepts that you didn't discuss in the rest of the paper.
How to structure the document
The only way to learn how to do this is to read other papers published in your field and emulate them. In general, remember to write each section independently and repeat the important statements. Readers probably won't go through your paper in order, but will instead jump from one section to another based on their interests. Don't assume that the reader knows what you said two pages earlier.
Target your audience
Your target audience should be very clear before you even start thinking of the outline. If you're writing a class report, your goal is to convince your professor you understood the topic and worked hard (so carefully explain the background and emphasize all the steps needed to finish the project).
If you're writing a paper for a general conference, then you should be aware that your audience is familiar with the topic and that they aren't experts. (Based on this, decide if a topic needs to be explained in more detail). If you're writing a paper targeted at a top conference in Software Defined Networking (SDN), you need to demonstrate that you're aware of all the related work published in that field. You can also remove the background section on SDN from your paper; your reviewers will be experts on that already.
How to know when you're done
It seems to me that an advisor's job is to keep offering advice for revisions up until the deadline. But this is the best thing that can happen to your writing, no matter how much it hurts to see the red comments every round of edits. When I go through the papers I wrote three years ago, I realize that all this feedback has improved my writing and made me a better writer.
The key to improvement
Trust those who are more experienced than you (advisor, professors, reviewers) and take their advice humbly. Pay attention to the writing at top conferences and published in top journals, and try to capture what they do that you’re missing.
Again, there are plenty of resources out there (start here) to help you with more specific questions. The suggestions I've featured in this post are those that have helped me the most during my grad school years and I hope some of them will also help you. Happy writing!